Lyse introduces a conversational interface to generate specs directly from the web app. You can now describe work in natural language through the Lyse Chat, refine details over multiple messages, and preview the generated task before publishing. Lyse pulls context from your connected tools to produce specs with real file paths, suggested assignees, and calibrated estimates. Responses stream in real time through live generation, so you see the spec taking shape as it's written. This release makes it faster to go from a rough idea to a complete, publishable spec, without leaving the Lyse dashboard.